Instrumentation aboard the latest generation geostationary satellites has allowed Earth observations that were previously not possible. The Geostationary Lightning Mapper (GLM) and Advanced Baseline Imager (ABI) aboard NOAA's GOES-R series, Advanced Meteorological Imager (AMI) on GEO-KOMPSAT-2A, and Advanced Himawari Imager (AHI) Himawari 8/9, is a MODIS-like instrument spectrally but can provide data at much higher temporal resolution suitable for monitoring fast changing processes in the Earth’s atmosphere and land surface. These capabilities were listed in the 2018 U.S. National Academies publication “Thriving on Our Changing Planet: A Decadal Strategy for Earth Observation from Space” as program of record that significantly complement the NASA’s own satellite programs. In this talk, the authors will report on the current status of NASA’s implementation of Earth Science Research from Operational Geostationary Satellite Systems (ESROGSS) program and how the funded projects will continue the success of NASA’s Earth Observing System (EOS) program.

New generation geostationary satellites provide very high temporal earth observation with multiple visible, near-infrared, and shortwave-infrared wavelength regions. Thus, these datasets can provide us a hyper-temporal resolution to monitor land surface. However, since geostationary satellite data are provided with top-of-atmosphere status, complex data pre-processing such as atmospheric correction, cloud masking are required. In this study, we report our status on its data processing and some key results.
First, cloud mask data are generated based on Yamamoto et al. (2018). We updated and simplified the algorithm, and conducted an intercomparison with MODIS products has also conducted. Our cloudmask outputs are basically consistent with MODIS ones. If we tested at tropical forest sites, we found that cloud cover ratio of land surface is overall consistent between our products and MODIS ones. Second, we estimated land surface reflectances using AHI. Our approach uses a commonly
used radiative transfer code, 6SV, and Himawari-8 aerosol products as atmospheric data. The estimated surface reflectances were evaluated with MODIS products and ground observation data. In the comparison with MODIS, we used two surface reflectance products, MOD09/MYD09 and MCD43. MCD43 is a BRDF corrected surface reflectance and provides its kernel function paramters. As a comparison of AHI and MODIS, we found that MODIS-based surface reflectances with adjustment to AHI observation geometry show the better agreement with those of AHI. Thus, matching the observation geometry is essential to evaluating surface reflectances. Furthermore, we confirmed that the effects of observation condition differences are much smaller when intercomparison is conducted using vegetation index.

Atmospheric 3D winds in the horizontal and vertical directions are critical for improving short-range and long-range forecasting. Such advancement in forecasting directly applies to research in a number of areas including convective processes, wildfire plumes and tornado prediction. Atmospheric Motion Vectors (AMVs) provide a passively sensed approach to quantifying horizontal motion and cloud heights, which are typically sourced from geostationary sensors due to the availability of high frequency observations. Recent work has shown that estimating AMVs by tracking individual pixels with dense optical flow is a promising new direction. In this work, we use a state-of-the-art convolutional neural network for optical flow (FlowNetS) in a physics-guided deep learning framework for predicting AMVs in the horizontal direction. The approach is semi-supervised and uses physically informed wind vectors from high-resolution numerical simulations (DYAMOND) for supervised learning followed by fine-tuning though warping and reconstruction of full-disk geostationary images (GOES-16). In the vertical direction, we use labels from the CALIPSO low-earth orbit satellite to predict cloud height from 16-band geostationary images with a neural network. We present results for both tasks on held-out time periods and secondary datasets.

Solar resource assessment is required for the feasibility study for the renewable energy power plants installation. Ground observation is considered as the most reliable datasets for the solar resource assessments but they are still limited due to the spatial resolution as well as their coverage. Therefore, solar irradiance derived from satellite imagery is useful for solar resource assessment, as well as climate change research without spatial limitation. The University of Arizona Solar Irradiance Based on Satellite (UASIBS) model was developed for the GOES-13 satellite to derive the downwelling surface solar irradiance over Arizona in the U.S. Then, it was modified to employ the satellite imagery taken by COMS over the Korean Peninsula as named into UASIBS - KIER (Korea Institute of Energy Research). Recently the Korean Government launched new satellite platform with advanced meteorological imager of which specification is similar to ABI and AHI onboard GOES-R and Himawari-8, respectively. The UASIBS - KIER model has been updated into version 2.0 to derive the solar irradiance at 500 m horizontal resolution. The preliminary results show that, for cloudy skies, the relative root mean square error values are 14.5% and 15.9% at the stations located in Korea and Japan, respectively. From 1995 to current time, satellite imagery from geostationary satellite can be accessible to build the solar resource map that includes the long-term variability. This presentaion will introduce the solar irradiance that is derived from UASIBS - KIER model with GMS-5, GOES-9, MTSAT-1R/2, COMS, GK-2A and Himawari-8 over the Korean Peninsula in upcoming conference. 

Surface urban heat island (SUHI) is one of the most significant human-induced alterations to the climate and can aggravate health risks for city dwellers during heatwaves. Geostationary satellites provide high-frequency land surface temperature (LST) observations throughout day and night, and offer unprecedented opportunities for exploring diurnal cycling of SUHI. We examined how the SUHI intensity varied over the course of the diurnal cycle in Boston using LST from the NOAA's latest generation of Geostationary Operational Environmental Satellites (GOES-R). We calculated the hourly SUHI intensity for both urban core and suburban areas. The maximum SUHI intensity for urban core occurred near noon (+3.0°C (12:00), +5.4°C (12:00), +4.9°C (11:00), and +3.7°C (12:00) in winter, spring, summer, and autumn, respectively). The maximum intensity for the suburb was about 3.0°C lower in spring and summer and 2.0°C lower in autumn and winter than that of the urban core. The minimum intensity occurred at nighttime, and ranged from −1.0°C to +1.0°C. The difference in nighttime intensity between urban core and suburb was insignificant for all seasons except summer. The SUHI intensity showed similar diurnal variations across seasons. Throughout the year, the maximum intensity (+2.7–+5.8°C) at urban core occurred at 11:00–14:00, while the minimum intensity (−0.6–+0.9°C) was commonly observed at 00:00–07:00 and 17:00–23:00. Relationships between SUHI intensity and drivers differed within a diurnal cycle, characterized by the strongest correlation with impervious surface area and population size during the middle of the day, and with tree cover at night. Our research highlights the great potential of geostationary satellites in revealing diurnal cycling of SUHI.

Land surface phenology (LSP) characterizes the seasonal dynamics of vegetation growth from time series satellite observations. It has been widely detected from polar-orbiting satellite sensors, such as MODIS and VIIRS. Because of persistent cloud cover in tropical regions, land surface observations are frequently missed from polar-orbiting satellites. Thus, vegetation phenology is either undetected or detected with high uncertainty. Fortunately, the Advanced Baseline Imager (ABI) onboard Operational Environmental Satellite (GOES) systems (GOES-16 launched in November 2016 and GOES-17 launched in March 2018) provides full disk (North and South America) observations every 15 minutes. It has a capability of obtaining 21-35 times more cloud-free observations than MODIS or VIIRS in Amazon rainforests. Therefore, this study detects LSP from ABI time series in the Amazon basin and comparing with the detections from VIIRS time series. Specifically, we calculate daily EVI2 (2 band enhanced vegetation index) from diurnal ABI observations after reducing the effects of Sun-satellite geometry on observations. The daily cloud-free ABI EVI2 observations are fitted with a hybrid piecewise logistic model to quantify temporal development of vegetation growth. From the established logistic model in each individual pixel, the curvature change rate is calculated to determine phenological transition dates that are greenup onset, maturity onset, senescence onset, and dormancy onset. The phenological dates from ABI EVI2 time series are further compared with the detections from VIIRS in 2018. The differences of phenological detections between ABI and VIIRS are distinguished in forests and croplands over the Amazon basin.  

The response of gross primary production (GPP) to moisture availability has been a source of uncertainty in modeling GPP. Periods of hydraulic stress from low soil moisture and/or high vapor pressure deficit can result in overestimation of GPP from light use efficiency (LUE) models. With the recent updates to geostationary satellites, we can now explore the diurnal cycles of GPP at a comparable spatial resolution and a much higher temporal resolution to polar-orbiting satellites. This is advantageous for studying the response of carbon uptake to moisture limitation and improving space-based GPP estimates, because the diurnal course of GPP and its diurnal coupling with water fluxes changes during times of soil moisture deficits. We estimated half-hourly GPP using multispectral data from the Geostationary Operational Environmental Satellites - R series (GOES-R) and half-hourly estimates of GPP from the Tonzi Ranch eddy covariance site in California, USA. The site is an oak-grass savanna in a Mediterannean climate which experiences seasonal summer droughts. We optimized parameters for three models to estimate GPP: a LUE model, a linear relationship between the near infrared reflectance of vegetation (NIRv) and tower GPP, and a light response curve (LRC). We used error summaries and diurnal centroids of GPP and latent heat flux to test whether GOES-R GPP estimates accurately captured summer drought conditions. We found that the LRC was the most successful in: (1) reducing bias in GOES-R GPP estimates compared to tower GPP, (2) replicating diurnal half hourly patterns during seasonal droughts, and (3) replicating diurnal patterns of carbon-water flux (de)coupling throughout the year. Our results can help develop diurnal estimates of GPP from geostationary satellites that are sensitive to fluctuating moisture conditions during the day and throughout the seasons.

Spectral vegetation index (VI) time series data have widely been used for characterizing terrestrial vegetation dynamics from regional to global scales. The utility of VI time series datasets is, however, often constrained by clouds, in particular, in cloud-prone tropical regions. While polar-orbiting satellite sensors such as Moderate Resolution Imaging Spectroradiometer (MODIS) typically acquire one or two day-time observations per day, this observation frequency has been found not enough to generate “clear-sky” VI time series data even at 16-day temporal resolution due to frequent cloud cover in humid tropical regions. A new generation of geostationary satellite sensors have been launched during the last decade, which are capable of imaging an Earth’s hemisphere at 10–15 min intervals and equipped with the spectral bands suitable for the derivation of VIs. Thus, these sensors can serve as another significant data source for the studies of vegetation dynamics in tropical regions. In this study, we investigated the degree of the temporal resolution improvement in the NDVI dataset by the use of Himawari-8 Advanced Himawari Imager (AHI), one of the third-generation geostationary satellite sensors, for Malaysian Borneo. AHI 10-min resolution NDVI data were extracted over Lambir Hills National Park, Sarawak, Malaysia for a period from July 2015 to December 2019. Terra MODIS NDVI data were also obtained for the same area and period. After the screening of suspicious observations, there remained only ~12 observations per year for Terra MODIS. In contrast, smooth NDVI temporal profiles were obtained from AHI by the 16-day maximum value compositing, approximately 23 observations per year. Comparisons of AHI NDVI with in situ cloud cover measurements indicated the validity of high NDVI values being least cloud-contaminated. These results indicate the advantage of Himawari-8 AHI for analysis of temporal vegetation dynamics for this cloud-prone tropical site.